Online Handwriting Recognition System for Ethiopic Characters: Analysis, Modeling and Design with Experimental Results

The popularity of handheld devices such as PDA has increased significantly. Mostly, such devices embed online handwriting recognition systems (OHRS), by which users¿ handwriting would be recognized and converted to digital form, implying the possibility of directly writing into a computer: a quite natural human-computer interaction. This book presents an OHRS for Ethiopic character set, which is the only native African writing system used for various languages in Ethiopia and Eritrea. A novel pattern representation technique has been designed and appropriate algorithms have been developed for the purpose of preprocessing, training and classification. Experiments have been conducted for 16 different writers and an average recognition accuracy of 97.4% has been achieved. This book should be useful to researchers on pattern recognition and particularly handwriting recognition systems. Researchers and/or developers who attempt to develop an OHRS for a discretely written character sets like the Ethiopic character set would benefit from the techniques used in the system described in this book. We strongly believe the same model can be used for another character set of similar nature.